融合PCA与混沌自适应遗传算法的图像识别
|
摘要
针对图像特征识别转为特征选择优化的问题,提出主成分分析与混沌自适应遗传算法结合的图像目标识别算法。首先通过PCA将图像特征线性组合转变为低维空间几个综合变量;同时改进遗传算法,利用混沌Tent模型生成均匀分布的初始种群、种群交叉及变异概率与种群适应度结合自适应变化,利用类内类间距与特征相关性重新构造适应度函数,采用精英保留策略进行子代选择,得到最优特征子集;最后利用概率神经网络与支持向量机分类器进行训练,识别测试图像。仿真实验表明,PCA与混沌自适应遗传算法结合能降低特征空间维数,使识别性能得到较好提升。
Aiming at the conversion of image feature recognition into feature selection and optimization,a method of image object recognition combined with principal component analysis and chaotic adaptive genetic algorithm is proposed. The algorithm firstly transforms the linear combination of image features into several synthetic variables in low-dimensional space through PCA. At the same time,it improves the genetic algorithm and uses the chaotic Tent model to generate evenly distributed initial populations. The population crossover and mutation probability with population fitness adapt to the change. The fitness function is reconstructed by using the intra-class and inter-class distance and feature correlations. The elite retention strategy is used to select children to obtain the optimal feature subset. The results are trained by using probabilistic neural networks and support vector machine classifiers to test image recognition. Simulation experiments show that the combination of PCA and chaos adaptive genetic algorithm can reduce the dimension of the feature space,and the recognition performance is improved.
Aiming at the conversion of image feature recognition into feature selection and optimization,a method of image object recognition combined with principal component analysis and chaotic adaptive genetic algorithm is proposed. The algorithm firstly transforms the linear combination of image features into several synthetic variables in low-dimensional space through PCA. At the same time,it improves the genetic algorithm and uses the chaotic Tent model to generate evenly distributed initial populations. The population crossover and mutation probability with population fitness adapt to the change. The fitness function is reconstructed by using the intra-class and inter-class distance and feature correlations. The elite retention strategy is used to select children to obtain the optimal feature subset. The results are trained by using probabilistic neural networks and support vector machine classifiers to test image recognition. Simulation experiments show that the combination of PCA and chaos adaptive genetic algorithm can reduce the dimension of the feature space,and the recognition performance is improved.
引文
[1]BHARDWAJ A,TIWARI A,BHARDWAJ H.A genetically optimized neural net-work model for multi-class classification[J].Expert Systems with Applications,2016,60(10):211-221.
[2]BUI K T T,BUI D T,ZOU J G,et al.A novel hybrid artificial intelligent approach based on neural fuzzy inference model and particles warm optimization for horizontal displacement modeling of hydropower dam[J].Neural Compution&Applications,2018,29(12):1495-1506.
[3]SU H Z,LI X,YANG B B,et al.Eavelet support vector machinebased prediction model of dam deformation[J].Mechanical Systems and Signal Processing,2018,110:412-427.
[4]WEI G W,ALSAADI F E,HAYAT T,et al.A linear assignment method for multiple criteria decision analysis with hesitant fuzzy sets based on fuzzy measure[J].International Journal of Fuzzy Systems,2017,19(3):607-614.
[5]魏弦.基于核主成分分析的热误差模型自变量优化[J].电子测量与仪器学报,2017,31(12):2017-2022.
[6]苑玮琦,于清澄.一种基于改进主成分分析的人脸识别方法[J].激光与红外,2007(5):478-480.
[7]涂昌慧,葛红,胡天亮.基于遗传算法和互信息公式结合的特征选择[J].华南师范大学学报:自然科学版,2014,46(6):28-32.
[8]申情,蒋云良,沈张果,等.基于组合混沌遗传算法的最小测试用例集生成[J].电信科学,2016,32(6):93-102.
[9]方仕勇,邹恩,辛建涛,等.新型混沌遗传算法在多约束QoS路由的应用[J].计算机应用研究,2012,29(8):3078-3080.
[10]王世玮,张迪,魏明磊,等.基于混沌多目标遗传算法的分布式电源规划[J].宁夏电力,2017(2):1-6+14.
[11]崔彦平,葛杏卫.复杂背景下回转体目标识别方法研究[J].半导体光电,2010,31(6):931-935+940.
[12]杨红梅,刘志刚,韩志伟,等.基于仿射不变矩的电气化铁路绝缘子片间夹杂异物检测[J].铁道学报,2013,35(4):30-36.
[13]宣国荣,郑俊翔,杨程云,等.巴氏距离和K-L变换结合的特征选择[J].计算机工程与应用,2004(36):90-92.
[14]唐莉,张永波,祝雪萍,等.基于主成分分析法与量变质变判别模式的汛期分期[J].水力发电,2018(5):27-31.
[15]李雪岩,李雪梅,李学伟,等.基于混沌映射的元胞遗传算法[J].模式识别与人工智能,2015,28(1):42-49.
[16]张大科,钱谦.一种新型自适应遗传算法在多峰函数优化中的应用[J/OL].软件导刊:1-4[2018-06-01].http://kns.cnki.net/kcms/detail/42.1671.TP.20180529.1755.010.html.
[17]金芬,孙春华,钟鸣.遗传算法中适应度函数的改进[J].机械设计与制造,2010(3):218-219.
[18]周爽,张钧萍,苏宝库.基于最速上升算法的超光谱图像波段选择搜索算法[J].计算机应用研究,2008(11):3501-3503.
[19]孙明华,崔海涛,温卫东.基于精英保留遗传算法的连续结构多约束拓扑优化[J].航空动力学报,2006(4):732-737.
[20]贺静,徐成武,任密林.基于概率神经网络的IPv6入侵检测技术研究[J].太原理工大学学报,2017,48(6):969-972+983.
[21]刘方园,王水花,张煜东.支持向量机模型与应用综述[J].计算机系统应用,2018,27(4):1-9.
[2]BUI K T T,BUI D T,ZOU J G,et al.A novel hybrid artificial intelligent approach based on neural fuzzy inference model and particles warm optimization for horizontal displacement modeling of hydropower dam[J].Neural Compution&Applications,2018,29(12):1495-1506.
[3]SU H Z,LI X,YANG B B,et al.Eavelet support vector machinebased prediction model of dam deformation[J].Mechanical Systems and Signal Processing,2018,110:412-427.
[4]WEI G W,ALSAADI F E,HAYAT T,et al.A linear assignment method for multiple criteria decision analysis with hesitant fuzzy sets based on fuzzy measure[J].International Journal of Fuzzy Systems,2017,19(3):607-614.
[5]魏弦.基于核主成分分析的热误差模型自变量优化[J].电子测量与仪器学报,2017,31(12):2017-2022.
[6]苑玮琦,于清澄.一种基于改进主成分分析的人脸识别方法[J].激光与红外,2007(5):478-480.
[7]涂昌慧,葛红,胡天亮.基于遗传算法和互信息公式结合的特征选择[J].华南师范大学学报:自然科学版,2014,46(6):28-32.
[8]申情,蒋云良,沈张果,等.基于组合混沌遗传算法的最小测试用例集生成[J].电信科学,2016,32(6):93-102.
[9]方仕勇,邹恩,辛建涛,等.新型混沌遗传算法在多约束QoS路由的应用[J].计算机应用研究,2012,29(8):3078-3080.
[10]王世玮,张迪,魏明磊,等.基于混沌多目标遗传算法的分布式电源规划[J].宁夏电力,2017(2):1-6+14.
[11]崔彦平,葛杏卫.复杂背景下回转体目标识别方法研究[J].半导体光电,2010,31(6):931-935+940.
[12]杨红梅,刘志刚,韩志伟,等.基于仿射不变矩的电气化铁路绝缘子片间夹杂异物检测[J].铁道学报,2013,35(4):30-36.
[13]宣国荣,郑俊翔,杨程云,等.巴氏距离和K-L变换结合的特征选择[J].计算机工程与应用,2004(36):90-92.
[14]唐莉,张永波,祝雪萍,等.基于主成分分析法与量变质变判别模式的汛期分期[J].水力发电,2018(5):27-31.
[15]李雪岩,李雪梅,李学伟,等.基于混沌映射的元胞遗传算法[J].模式识别与人工智能,2015,28(1):42-49.
[16]张大科,钱谦.一种新型自适应遗传算法在多峰函数优化中的应用[J/OL].软件导刊:1-4[2018-06-01].http://kns.cnki.net/kcms/detail/42.1671.TP.20180529.1755.010.html.
[17]金芬,孙春华,钟鸣.遗传算法中适应度函数的改进[J].机械设计与制造,2010(3):218-219.
[18]周爽,张钧萍,苏宝库.基于最速上升算法的超光谱图像波段选择搜索算法[J].计算机应用研究,2008(11):3501-3503.
[19]孙明华,崔海涛,温卫东.基于精英保留遗传算法的连续结构多约束拓扑优化[J].航空动力学报,2006(4):732-737.
[20]贺静,徐成武,任密林.基于概率神经网络的IPv6入侵检测技术研究[J].太原理工大学学报,2017,48(6):969-972+983.
[21]刘方园,王水花,张煜东.支持向量机模型与应用综述[J].计算机系统应用,2018,27(4):1-9.